| To date,the most common tools and techniques applied to healthcare forecasting(whether clinical or non-clinical)includes the mathematical methods like likelihood test,seasonal auto regressive integrated moving average,o t h e r time series regressions,Cox hazard models,exponential smoothing,etc.Others include Kalman filtering,linear prediction,trend estimation,growth curve.Despite the successful application of these traditional statistical models in healthcare decision making,the complexity of the human body,the multidimensional and non-linear nature of clinical characteristics and healthcare systems limits their predictive ability hence a challenge on the effective planning and service quality delivery.With the emergence of data mining and machine learning techniques,Artificial Neural Networks,group method of data handling,support vector machines,etc has been experimented to support healthcare forecasting scenarios such as hospital demand,diagnosis and prognosis and predicting more complex healthcare decision scenarios(forecasting demand and service quality)with high accuracy over the conventional statistical models albeit their weaknesses.However,judgmental forecasting model are among the less frequently used forecasting models when it comes to healthcare decision making.Within the last decade,medical forecasting literature has seen significant attempt to revisit the role of clinician(doctors)judgment in medical decision making as a complement of Evidence Based Medicine(EBM)due to its practical limitations.This study attempts to determine the degree to which wisdom of the crowd tools such as “prediction market” is applicable within the healthcare market.Firstly the prediction market technique was used to predict patient flow in the emergency departments of the affiliated hospital of the Jiangsu University and the affiliated hospital of the Guilin Medical University in the Guanxi province.The results were then compared to the outcome of similar predictions using the exponential smoothing method,seasonal autoregressive moving averages method,time series regression method and feedforward artificial neural network method.Secondly the outcome of predicting the survivability of patient(clinical forecasting)using the prediction market technique is compared with similar predictions by the Cox hazard proportion regression model and feed-forward artificial neural network.Overall the post-sample mean absolute percentage error of prediction based on the different forecasting models exceeded the post sample prediction error using the prediction market technique in both clinical and non-clinical conditions forecasting scenarios.However,an ANOVA test results and a Tukey test of post-hoc multiple comparison failed to establish any statistical difference in the observed mean absolute percentage errors.This information provides a strong case to complement existing medical forecasting techniques with other models such as prediction market in both clinical and non clinical forecasting of health services.Generally,it seems that medical decision making(especially forecasting,diagnosis,therapy,prognosis,communication etc)can benefit from crowd wisdom for the temporal accumulation of medical information over time which may lead to the development of a ‘Swarm Intelligence’ algorithm where pieces of information are brought together to form a part of the ‘Swarm’ to stimulate intelligent informed behaviors in medical decision making.It can act in a coordinated manner despite the lack of leadership or an external controller.Many examples can be seen in the nature of swarms that perform some collective behavior such as the ant colony,etc without any individual who controls the group,or to be aware of the overall behavior of the group.In these swarms,each individual has a stochastic behavior that depends on its local perception of the community hence possible to design a system of swarm intelligence that is scalable(maintain its function,while increasing its size without the need to redefine how its parts interact),parallel and fault tolerant.The above concepts of swarm intelligence are already inspiring new initiatives in medical literature and practice such as the online medical forum by the Indian Orthopedic Research Group(IORG)and similar ones in other parts of the world(Sato et al.,2005).In these forum surgeons presents the clinical and radiological details of their cases to elicit comments from other clinicians based on their personal experiences and familiarization of the current literature on the subject.This helps clinicians to obtain different perspectives on a variety of topical issues affecting their practice by quickly sharing knowledge and effectively using ‘Wisdom of the medical Crowds.The “Journal of Orthopedic Complications” and the “Orthopedic Case Bank” have also been launched by the Indian Orthopedic Research Group(IORG)to accept only complications or complicated cases to elicit discussions by the community of orthopedic surgeons.With time this “Bank” can grow and become a warehouse with a variety of cases that can be grouped together and searched simultaneously by individual clinicians and others who need them.The next step is to regularize the forums and develop a good publishing format and start publishing these rich case discussions,either a part of a journal or in other citable online format in public domains.This will make this information available to more viewers and also to generations to come as a template of current thought process.Algorithms can be developed based on case characteristics to find the nearest neighbor and also to provide recommendation based on data in the ‘Bank’... |
PDF Full Text Request